a black box on a PDF is not redaction

A PDF is not a picture of a document; it is a list of instructions for drawing one. Text is stored as text, with the font and the position to put it in. A black rectangle is one more instruction, executed after the text, so it appears on top of it — the way a sticker sits on a page rather than the way ink removes what was there.

Nothing about that instruction touches the text. Select the area in any reader and the words come out. Copy them, or run any of the extractors built into every operating system, and there they are. The demonstration above uses the same one your browser already has.

It is an easy mistake because the file looks right. You open it, you see black boxes, and the check that would reveal the problem is one nobody thinks to run on their own work.

Court filings have been published with the names of protected witnesses under black boxes that any reader could lift. Government reports have gone out with redacted passages that journalists recovered the same afternoon. Corporate filings have disclosed settlement figures the same way. In every case the file was reviewed, approved and released by people being careful — careful about the wrong thing.

The pattern is always the same: the tool made a mark that looked like a redaction, and nobody read the file back to check.

The text has to stop being in the file. In practice there are two honest ways to do that. One is to edit the page's instructions and delete the glyphs under the box — precise, and unforgiving: a mistake in the arithmetic leaves three words behind, and you will not know. The other is to redraw the marked page as an image of itself with the box painted in, which cannot leave text behind because the page no longer has any.

The redaction tool here takes the second route, on the marked pages only, so the rest of the document stays searchable. Then it does the thing nobody does: it reopens its own output, extracts the text, and tells you which words are gone.

Whatever tool you use — this one or another — read the result back before you send it. It takes ten seconds and it is the only check that matters.

Removing the text is not the whole job. A PDF also carries metadata — the author, the software, sometimes the original filename and the editing history — and none of that is under your black box. A document redacted perfectly can still be traced through the name of the person whose copy of Word produced it.